I automated my own job search

A service that scores vacancies against my profile, tracks my application funnel and shows which skills are missing most often.

Why

Rejections feel bad when you look at them one by one. I wanted to see the whole picture: how many applications, where they stop, and which skills the market asks for most often.

What it does

A vacancy score from 0 to 100 with explanations: commercial experience counts fully, lab experience counts partly; plus salary, labour contract, remote work or commute, night shifts, “talent pool” vacancies. Every point is explained: “+15 hybrid”, “−30 office only, 90 min commute”.
A funnel: application → HR → interview → offer or rejection, with dates and notes, and a next step with a date.
Statistics: applications this week against a target, response rate, the stage where rejections happen, the top missing skills, follow-up reminders.
Metrics in Prometheus and a Grafana dashboard.

How it works

A Go service with PostgreSQL in my Kubernetes cluster. Postgres is a StatefulSet on the official image, secrets are SealedSecrets, the image is distroless and runs as non-root. CI: tests, Semgrep, govulncheck, Trivy; deployment with ArgoCD.

I wrote the code together with an AI assistant (Claude Code). My part was the idea, the scoring rules, the requirements, the infrastructure, the deployment and operations.

What I learned

Data helps with anxiety: when you see that rejections come at the CV stage, it’s clear what to fix — the CV and the keywords, not yourself.